Experience the Magic of Applying Science in Transforming Organizations

We introduce advanced AI simulation and predictive frameworks, integrated with LLM agentic workflows and immersive virtual environments. We present physics-informed and graph transformer neural operators for real-time predictions in manufacturing setting. These technologies, along with an agentic digital twin for automated design optimization, are demonstrated within NVIDIA Omniverse for interactive visualization, showcasing a unified approach to high-fidelity simulation and decision support.

The next wave of AI isn't just about data; it's about creating and simulating reality. In this session, our experts from NVIDIA and Quantiphi will provide an exclusive look at how we're using powerful tools like NVIDIA Earth-2 and PhysicsNeMo to build the next generation of AI-powered simulations. We’ll share insights from two real-world projects: forecasting hurricane risk for an insurance client and an advanced digital twin for aerospace manufacturing. This discussion will explore the "why" and "how" behind these technologies, demonstrating their ability to offer unparalleled accuracy and efficiency. You will gain a clear understanding of the innovative applications and tangible business value of AI-driven simulation and digital twins.

This session presents advanced optimal workforce scheduling research grounded in real-world industrial applications, addressing customer care, operational efficiency, and employee satisfaction. We showcase our sophisticated genetic algorithm through case studies in nurse staffing and retail personnel allocation, highlighting quantifiable improvements and addressing complex constraints. Our ongoing research integrates AI, including graph neural networks, reinforcement learning, and generative AI, for dynamic and adaptive scheduling solutions, with a demonstration of our nurse assignment scheduler.

Discover a cutting-edge framework for analyzing warehouse simulation data, combining Knowledge Graphs with LLM-based agents for automated inefficiency diagnosis. This approach transforms unstructured DES output into a semantically rich KG, enabling an LLM agent to perform iterative, self-correcting reasoning on natural language queries. We demonstrate its superior accuracy in identifying bottlenecks and its advanced diagnostic capabilities for complex operational issues.

Drug design is a complex multi-stage process, the lead candidates that are designed can fail at different stages making the whole process a time and resource intensive process where a handful of lead candidates make it to clinical trials. Using generative models for molecule design is seeing a lot of interest with different generative models tackling the design problem from different angles. Our team has designed a multi-objective approach with an ensemble of generative models and flexible objective functions that can be tuned to design novel scaffolds, redesign known scaffolds, optimize the scaffolds for different properties. This multi-objective approach is flexible so that the chemists can use it in their process to combine expert knowledge with the sampling and scale capabilities of AI.

Proteins are the intricate molecular machinery orchestrating every biological process. From muscle contractions to immune responses, their astonishing precision makes them indispensable tools. This fundamental importance was highlighted by the Nobel Prize recognition in 2024 for advances in protein structure and design. Understanding the function of proteins and their behavior is a key research area. Proteins are also studied to optimize their properties for their use in industrial and therapeutic applications. One such property is thermostability, a protein’s ability to withstand higher temperatures make it a favorable candidate in industrial applications. With the rise in the use of generative models in biology, protein design using AI is an exciting area of research.

Neoantigens/neoepitopes, have particularly garnered a lot of interest in markets as they’re very promising therapeutic targets for a host of precision oncology therapies. As cancer cells go through mutations, these mutations lead to proteins that can be precisely targeted. Neoepitopes are new, "mutated" ID badges that appear specifically on the surface of cancer cells but are NOT present on healthy cells. Creating a scalable approach that is cloud agnostic and can run fast is an attractive option to identify neoantigens from whole genome data, this makes precision therapies for cancer a reality. Specialized models are needed to identify mutations and predict which of these mutations can be targeted for precision therapy.